EscapistsReintroduced/docs/evidence/coordinated-ai-2026-10-13/policy-final-targeted.log

PASS: tests/test_ai_pursuit_replay.py
.............
----------------------------------------------------------------------
Ran 13 tests in 0.689s

OK

PASS: tests/test_hq_ammo_dispatch.py
PASS: new/retasked search and radio delivery ammo gates; active tasks and delivered release preserved

PASS: tests/test_hq_ammo_reporting.py
PASS: actual carried/mounted early ammunition policy and status-first shared uplink; complete threshold, phases, cooldown, failures requested check-ins and non-urgent detached carried-mine status

PASS: tests/test_hq_ammo_sampler.py
PASS: actual own-inventory ammo inspection deduplicates loaded/carried/shared wells, separates chambers/launchers, freezes scalar snapshots and treats unsupported/capped/foreign inventory as unknown

PASS: tests/test_hq_assignment_lifecycle.py
PASS: actual failed-retask deadline rollback plus assignment cancellation/release preserves radio ownership, native waypoint lifecycle and failed-order retry boundaries

PASS: tests/test_hq_body_observer.py
PASS: actual body observer enforces owned physical roster, local death-only candidates, native trace gating, budget/cadence/caps, immutable body-only snapshots, radio retries and confirmed suppression

PASS: tests/test_hq_convoy_combat.py
HQ convoy delivered-report planning and downlink regressions passed

PASS: tests/test_hq_convoy_dispatch.py
PASS: actual authenticated convoy dispatch, active/ready admission gates, canonical receipt, safe unconsumed preflight retry, pool refusal, retained partial failure and consumed no-retry

PASS: tests/test_hq_coordinated.py
PASS: actual coordinated adapter report gates, local evidence, delivered-force snapshots, ownership/lease checks, continuous-contact arbitration, legacy isolation and shadow no-native-order boundary

PASS: tests/test_hq_coverage.py
PASS: actual bounded delivered-negative coverage, replay/freshness/expiry/invalid guards, oldest eviction, small-circle and nearest-unswept selection, reserved-cell exclusion and cleanup

PASS: tests/test_hq_delivered_casualties.py
PASS: actual delivered slot-union bodies preserve overlap/disjoint evidence, replay idempotence, status subset integrity, full-wipe closure and formation-local bounds

PASS: tests/test_hq_delivered_task.py
PASS: actual delivered-task lineage, expiry, old-order retention, native commit, release, road precision expiry and duplicate delivery

PASS: tests/test_hq_doctrine.py
PASS: actual objective heading/threat/approach rules, persistent doctrine weight, fresh-contact coverage invalidation, stale-sweep rejection and atomic radio retask ownership

PASS: tests/test_hq_execution.py
PASS: actual execution lifecycle, original physical arrivals, failed movement, quiet footprint, local support expiry and delivered report acknowledgments
PASS: native support wiring requires own activity, soldier suppression, actual projectile event and immutable roster; no waypoint-completion arrival shortcut
PASS: actual finite native LOD leases restore prior overrides and protect transferred agents, player ownership and intervening overrides

PASS: tests/test_hq_hunter_operations.py
PASS: actual hunter reservation ledger, pending/delivered replacement, independent six/eight caps, stale/empty status, finite observation deadline, no hidden force-position lookup

PASS: tests/test_hq_hypotheses.py
PASS: actual five/six bounded hypotheses, observation-relative ETA, future lead, hiding/reverse, local negative weighting, immutable report lineage, road bends/heights and finite cache budgets

PASS: tests/test_hq_local_awareness.py
PASS: captured sighting retention, immutable details, expiry, replay rejection, bounded sound uncertainty/cooldown, radio loss, priority and overdue-status fairness

PASS: tests/test_hq_local_coverage.py
PASS: actual10/40m local negative patches, same-cell hidden-ground preservation, proportional residual belief, replay, chronology, physical-report guards, caps and cleanup

PASS: tests/test_hq_local_handoff.py
PASS: actual original-roster local handoff,100m3D bound,busy carrier/budget exclusion,origin interruption including deletedsource,normal radio delivery and per-source replay history

PASS: tests/test_hq_local_handoff_intake.py
PASS: actual visual intake retains source/time/position/id, rejects replay, prefers usable own route, excludes simultaneous callers and conservatively resets negative-sweep dwell

PASS: tests/test_hq_memory.py
PASS: actual HQ memory rules/service preserve delivered snapshots, chronology, replay bounds, decay, motion limits, identity and negative-search revisions

PASS: tests/test_hq_mine_sampler.py
PASS: actual carried PMN4 inventory, original-owner guards, bounded deduplication, independent unknown state and detached counts

PASS: tests/test_hq_mounted_ammo_sampler.py
PASS: actual mounted hierarchy sampling, ownership exclusion, deduplication, completeness/caps, quarter threshold and detached validation; no runtime policy or ammunition mutation
PASS: WORKBENCH opt-in diagnostics distinguish cycles/cap/storage/manager/mount failures without repeated native traversal

PASS: tests/test_hq_mounted_binding.py
PASS: actual HQ original convoy-root admission rejects wrong force/server/group and never supplies replacement lookup

PASS: tests/test_hq_mounted_dispatch.py
actual mounted HQ dispatch planning, radio, ownership and lifecycle checks passed

PASS: tests/test_hq_mounted_intercept.py
PASS: actual mounted intercept eligibility/order chronology, fixed destinations, native road selection bounds, width, segment heading, clearance and settlement delegation; no native driving claim

PASS: tests/test_hq_mounted_intercept_pair.py
PASS: actual paired arc geometry,40m/80m spacing, separate original polylines, both-position validation and128road/4096segment budget; no native driving proof

PASS: tests/test_hq_multiconvoy_supply.py
PASS: actual HQ rotates finite donors, skips unavailable/contacted trucks, uses delivered reports and dispatches at most one trip per planning cadence

PASS: tests/test_hq_operations.py
PASS: actual operation selection/replay, pending reservations, detached reports, identity/sequence/time gates, terminal lifecycle, bounded ledger, finite roles, no shadow mutation or hidden-player inputs

PASS: tests/test_hq_planner_budget.py
PASS: actual full planner selection/commit body, 360 equivalence scenarios, orders/reservations/cooldowns identical; one128-unit scan vs >10000 repeated unit visits; no FPS claim

PASS: tests/test_hq_pursuit.py
PASS: actual bounded pursuit roles/objectives/scores, stale search degradation, physical/busy/home guards, finite lifetime and two-bearing escape gap; no hidden player reads or spawning
PASS: actual finite45..110 radius and both ordinary search queue snapshot expressions

PASS: tests/test_hq_radio_node.py
PASS: actual radio node rejects current incapacitation, death, ownership and faction changes despite cached inventory; one-second capability refresh remains explicit

PASS: tests/test_hq_report_selection.py
PASS: actual per-squad bounded report selector rotates visible targets, survives ordering/turnover, records only queued attempts, and preserves radio intake ordering

PASS: tests/test_hq_road_intercept.py
PASS: actual delivered road eligibility, ahead projection without chords, reservations, stock road query bounds, terrain/water/obstruction refusal and clean outputs

PASS: tests/test_hq_status_sampler.py
PASS: own initial roster, explicit retained corpse evidence, unconscious/missing/ref transfer guards, fixed strength, no replacement adoption, detached samples, surviving radio sender, immutable convoy root/ledger, mounted capture release and lost-root unknown

PASS: tests/test_hq_task_reservations.py
PASS: actual HQ active/pending reservations, interruption, committed replacement, bounded coverage and unchanged force counts

PASS: tests/test_hq_uplink_fairness.py
PASS: actual report methods reproduce old starvation; status-first scheduler delivers check-ins and requires 30s without combat or actual visible contacts before negative reports

PASS: tests/test_hq_vehicle_seats.py
PASS: production mounted structural seat scan ignores >128 carried/cargo descendants, shares exact native mount/magazine ownership, preserves attachment seats and exact occupants, rejects unknown/cyclic/oversize structures; native binding still requires qualification.

PASS: tests/test_hq_watch.py
PASS: delivered-only status snapshots, two missed check-ins/reply deadline, last reported coordinates, episode/replay guards, confirmed-loss idempotence, detached/validated mounted payload replacement 256-record cap and detached carried-mine radio delivery/chronology

PASS: tests/test_hunter_capture.py
PASS: actual player-entry hook latches original hunter capture before native callbacks; both release layers cancel ownership without new entities/refund; bounded current-owner guards and prior-release policy

PASS: tests/test_hunter_close_contact.py
PASS: close observed threat, low-speed boundary, invalid samples, and all 64 original-crew/gunner/player eligibility combinations

PASS: tests/test_hunter_combat.py
PASS: only recent native detected/identified threats suspend hunter travel; stale/disarmed/destroyed contacts cannot strand patrols
PASS: actual hunter combat transition cancels only its travel once, holds through renewed contact and resumes after quiet; unregistered startup preserved

PASS: tests/test_hunter_coordination.py
PASS: actual HQ allocator selects reachable reports, distinct cover/flanks, simultaneous fronts, bounded reservations and failed-radio/placement rollback

PASS: tests/test_hunter_driver_handoff.py
PASS: accept a fit original crew member assigned by native boarding; reject outsiders, players, casualties, wrong group and missing compartments

PASS: tests/test_hunter_fallback.py
PASS: exhausted/blocked opening keeps configured radius; distant objective fallback rejected; bounds, player clearance, separation, force cap and retry delay retained

PASS: tests/test_hunter_gun_team.py
PASS: only original fit loaded healthy uncaptured gunners stay mounted; driver can leave; infantry readiness excludes the retained gunner but regroup requires all survivors

PASS: tests/test_hunter_lifecycle.py
PASS: actual hunter release preserves survivors/HQ, releases driving ownership once, delegates capture/teardown, and excludes prison/patrol/mortar from generic objective defence
PASS: actual hunter cancellation releases exactly the pending source/message once; teardown clears ownership without HQ/game/world

PASS: tests/test_hunter_opening.py
PASS: real opening road geometry enforces wide azimuths; fresh delivered contact preempts patrol cooldown, respects radio failures and does not starve in-flight interception

PASS: tests/test_hunter_profiles.py
PASS: missing/invalid faction crew defers before any native spawn or allowance insertion; all registered RHS factions have explicit soldier and medic profiles
PASS: vehicle radio issued once; pending, refusal and looting cannot duplicate or fabricate a physical radio link
PASS: actual vehicle-radio link requires the original seated crew, physical manpack and live uncaptured mission vehicle; looting, dismount, release and teardown remove it

PASS: tests/test_hunter_reserve.py
PASS: reserve uses HQ reports/dispatch deadlines, preserves radio uncertainty, spends once, respects cooldown/availability/three-reserve/15-vehicle caps

PASS: tests/test_hunter_search_memory.py
PASS: delivered ground search survives the 60-second mortar cutoff, with five-minute/500m uncertainty bounds and no live target reads; mortar policy unchanged

PASS: tests/test_hunter_sector_search.py
PASS: searches expire ten minutes after observation, ignore replays, wait at arrival, widen within 1 km and bound road queries

PASS: tests/test_hunter_tactics.py
PASS: separated rear vehicle approaches, opposite infantry sectors, radio-gated ownership, failed-order preservation, visual cover or bounded supported reconnaissance, bounded dismount/search/regroup and loss cleanup
PASS: production dismount/boarding allocation, all crew allowances, waypoint ownership, one-time vehicle detach/restore and capture/teardown guards
PASS: covering support requires the original conscious turret gunner, valid mounted driver, live crew and delivered assignment
PASS: alternate positions exclude failed and occupied stages, roll back on failed planning/allocation, and preserve the total time budget

PASS: tests/test_hunter_transport.py
PASS: finite 2-UAZ/1-truck rotation, exact physical crew seats, one medic, and no smaller/armed substitute for transport

FINAL TARGETED RESULT: 54 scripts, 0 failures